Report Abuse

Basic Information

This repository contains multiple working agent examples implemented across different programming languages and frameworks so developers can compare implementations side-by-side and use them as templates to start new projects. Examples focus on common LLM agent patterns such as a Customer Support Agent and a Vibe Coding Agent and are provided in both Python and TypeScript. Framework-specific examples include Agno, DSPy, Google ADK, InspectAI, LangGraph, Letta, Pydantic AI, smolagents, Ax, Inngest AgentKit, LangGraph.js, Mastra and a zero-dependency Cloudflare Worker. Each example is self-contained with code and tests, some include UI demos or trace links, and all examples use the same Scenario tests and the gemini-2.5-flash-preview-04-17 model for verification. The README documents cloning, installing, running tests and entering debug mode and invites contributors to add frameworks and use cases.

Links

App Details

Features
Side-by-side, framework-specific example projects in Python and TypeScript that implement identical agent use cases to facilitate direct comparison. Self-contained example folders include source code, tests, and shared utilities that simulate external systems so examples remain readable and portable. Test suites are provided (pytest for Python, npm test for TypeScript) and examples support a debug mode that allows interactive chatting with the agent. Some examples include UI files or trace links for observability. All examples run the same Scenario verification tests and target the same model to ensure consistent behavior. Contribution guidance and a checklist of additional workflows and frameworks to add are included to grow coverage.
Use Cases
This repo helps developers evaluate and bootstrap LLM agent projects by providing ready-made, consistent examples across many popular agent frameworks. It reduces experimentation overhead by standardizing prompts, tests and simulated external services so implementers can focus on framework differences rather than building scaffolding. The included tests and use of a single model for verification make behavior comparisons reproducible. Debug modes and sample UIs enable quick iteration and inspection. Contribution instructions and templates make it easier to add new frameworks or use cases, accelerating learning and adoption of agent patterns and simplifying the process of starting a new agent-based project.

Please fill the required fields*